How can I use the eig function

You can use lambda_min, lambda_max, lambda_sum_largest, and lambda_sum_smallest.

I am sorry that my English is poor and I start to use CVX recently.I did not get your point.The problem is as following:

Is this problem convex?And can I use the A==-hermitian_semidefinite(N) denotes the constraint A is negative semidefinite?

I believe that will work. Or use

`-A == hermitian_semidefinite(N)`

Note that as I guessed might be the case in your other question, you want to constrain a matrix to be negative semidefinite. That is not the same as constraining a matrix to not be positive semidefinite. They are not equivalent because the latter includes the possibility of the matrix being indefinite.

I understand it with your clear explanation!Many thanks!